pycopyql 0.4.0

Exports a subset of data from a relational database

pycopyql

Exports a subset of data from a relational database.

Inspired by copyql. ❤️

Installation

Use pip.

pip install pycopyql

Usage

pycopyql [-h] [-c CONFIG] [-f FORMAT] connection output query [query ...]

• -h: Shows usage information
• -c: Path to a configuration file, defaults to ./pycopyql.py
• -f: Output format, sql and json are supported, defaults to sql
• connection: Name of the connection to use, defined in the configuration file
• output: Path to output file
• query: One or more query strings (delimited by spaces) of the format TABLE.COLUMN:VALUE (e.g. accounts.id:1)

Example:

pycopyql -c ./path/to/pycopyql.py -f json my_connection path/to/output.sql accounts.id:1 users.id:2

Configuration

pycopyql uses a Python file for configuration. It looks like this.

# pycopyql.py

config = {
    'connections': {
        'my_connection': {
            'drivername': 'mysql+mysqlconnector',
            'username': 'myuser',
            'password': 'mypassword',
            'host': 'localhost',
            'port': 3306,
            'database': 'mydatabase',
        }
    }
}

The file must define a config variable that is assigned a dictionary. That dictionary must have a 'connections' key that is likewise assigned a dictionary.

Within 'connections', each key-value pair represents the configuration for a single database connection. The key is a name used to reference the connection when invoking pycopyql.

Under the hood, pycopyql uses the popular SQLAlchemy library. Most settings for each connection correspond to the settings used for SQLAlchemy URLs.

Resolvers

A resolver in pycopyql is a function that receives the names of a table and a column contained in that table and returns a list of table-column pairs related to the specified table-column pair that pycopyql will then search for data to include in its output.

Let's take a look at an example of a resolver that's bundled with pycopyql.

❯ python3
Python 3.7.0 (default, Jun 29 2018, 20:14:27)
[Clang 9.0.0 (clang-900.0.39.2)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from pycopyql.resolver.inflector import dependents_inflector
>>> help(dependents_inflector)

Help on function dependents_inflector in module pycopyql.resolver.inflector:

dependents_inflector(meta, table, column)
    Uses inflection of column names to return foreign keys from tables that reference the specified column if it is a primary key.

    Parameters:
        meta (sqlalchemy.schema.MetaData): Metadata for the database structure
        table (string): Name of the table containing the column being examined
        column (string): Name of the column being examined

    Returns:
        list: List of dictionaries each containing the table and column of foreign key columns that reference the specified column

Let's assume that we're using this resolver with a database that looks like this.

+-----------+     +------------+     +---------+
| libraries <--+  | books      |  +--> authors |
+-----------+  |  +------------+  |  +---------+
| id        |  |  | id         |  |  | id      |
| name      |  +--+ library_id |  |  | name    |
| owner     |     | title      |  |  +---------+
+-----------+     | author_id  +--+
                  +------------+

Given a table name of 'libraries' and a column name of 'id', dependents_inflector would return this.

[
    { 'table': 'books', 'column': 'library_id' },
]

pycopyql.resolver.inflector also includes another resolver called dependencies_inflector, which works in the opposite direction of dependents_inflector. That is, where dependents_inflector returns foreign keys that reference a primary key, dependencies_inflector returns the primary key corresponding to a foreign key.

Using the above database as an example again, given a table name of 'books' and a column name of 'library_id', dependencies_inflector would return this.

[
    { 'table': 'libraries', 'column': 'id' },
]

A custom resolver that handles both of these cases using the bundled resolvers shown above might look like this.

# pycopyql.py

from pycopyql.resolver.inflector import dependents_inflector, dependencies_inflector

def my_resolver(meta, table, column):
    dependents = dependents_inflector(meta, table, column)
    dependencies = dependencies_inflector(meta, table, column)
    return dependents + dependencies

config = {
    'connections': {
        'my_connection': {
            # SQLAlchemy connection parameters go here

            'resolver': my_resolver,
        },
    },
}

In fact, pycopyql.resolver.inflector contains yet another resolver called inflector that functions exactly this way.

Custom resolvers can operate in many ways. For example, one might use foreign keys to determine relationships. The dependents_foreign_key and dependencies_foreign_key resolvers contained in pycopyql.resolver.foreign_key are equivalent to the dependents_inflector and dependencies_inflector resolvers described earlier. pycopyql.resolver.foreign_key also includes the foreign_key resolver, which is likewise equivalent to the inflector resolver described earlier.

For some cases where database column naming conventions are inconsistent, or where one table requires multiple references to another table (e.g. identifiers of the last users who created and updated a record), a resolver can use a static list of keys specific to that database. Here's an example using the database schema shown earlier.

def my_resolver(meta, table, column):
    keys = {
        'libraries': {
            'id': [ { 'table': 'books', 'column': 'library_id' } ],
        },
        'books': {
            'library_id': [ { 'table': 'libraries', 'column': 'id' } ],
            'author_id': [ { 'table': 'authors', 'column': 'id' } ]
        },
        'authors': {
            'id': [ { 'table': 'books', 'column': 'author_id' } ],
        },
    }
    if table in keys and column in keys[table]:
        return keys[table][column]
    return []

pycopyql.resolver.static contains a convenience function, static_map, that takes in a dictionary like the one assigned to keys in the example above and returns a resolver that will use that static map in the same way.

from pycopyql.resolver.static import static_map

keys = {
    'table_1': {
        'column_1': [
            { 'table': 'other_table_1', 'column': 'other_column_1' },
            { 'table': 'other_table_2', 'column': 'other_column_2' },
            # ...
        ]
    },
    'table_2': {
        # ...
    },
    # ...
}

config = {
    'connections': {
        'my_connection': {
            # ...

            'resolver': static_map(keys),
        },
    },
}

Custom resolvers can use several of the resolvers described here together simultaneously. For example, if you wish to use inflection, but have some cases for which you must provide a static mapping, you can do both using a custom resolver.

from pycopyql.resolver.inflector import inflector
from pycopyql.resolver.static import static_map

static_keys = {
    # ...
}

static_resolver = static_map(static_keys)

def my_resolver(meta, table, column):
    static = static_resolver(meta, table, column)
    inflected = inflector(meta, table, column)
    return static + inflected

Output Formats

By default, pycopyql supports sql and json output formats. You can add additional formats or replace the bundled export functions for supported formats with your own.

In your configuration file, on the same level as the 'connections' key, add an 'exporters' key. The corresponding value should be a dictionary keyed by a format name that will be passed to pycopyql using its -f flag. The value should be a function that outputs given data in that format.

Let's take a look at one of the bundled export functions.

❯ python3
Python 3.7.0 (default, Jun 29 2018, 20:14:27)
[Clang 9.0.0 (clang-900.0.39.2)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> from pycopyql.export import export_sql
>>> help(export_sql)

Help on function export_sql in module pycopyql.export:

export_sql(meta, data, output)
    Outputs data as SQL INSERT statements.

    Parameters:
        meta (sqlalchemy.schema.MetaData): Metadata for the database structure
        data (dict): Dictionary keyed by table name of dictionaries corresponding to table rows
        output (string): Path for output file

    Returns:
        None

If you wanted to override the bundled SQL exporter, here's what it might look like.

def my_sql_exporter(meta, data, output):
    # ...

config = {
    'connections': {
        # ...
    },
    'exporters': {
        'sql': my_sql_exporter,        
    },
}

License

MIT License